Moving directory tree one level higher.

doc/README

@@ -1,3 +1,27 @@
+                                   ADDA 0.77
+                                   *********
+                                "Amsterdam DDA"
+
+                    Maxim A. Yurkin(1,2) and Alfons G. Hoekstra(1)
+
+             (1) Faculty of Science, Section Computational Science,
+                       of the University of Amsterdam,
+              Kruislaan 403, 1098 SJ, Amsterdam, The Netherlands,
+                   tel: +31-20-525-7530, fax: +31-20-525-7490
+
+             (2) Institute of Chemical Kinetics and Combustion,
+               Siberian Branch of the Russian Academy of Sciences,
+                  Institutskaya 3, Novosibirsk 630090 Russia,
+                    tel: +7-383-333-3240, fax: +7-383-334-2350
+
+                           email: [email protected]
+
+                          last revised: 4 April 2007
+
+                Copyright (C) 2006-2007 University of Amsterdam
+        This software package is covered by the GNU General Public License.
+
+
                 ##             ##### ##         ##### ##         ##
              /####          /#####  /##      /#####  /##      /####
             /  ###        //    /  / ###   //    /  / ###    /  ###
@@ -17,31 +41,6 @@
          ##               ##               ##             ##
 
 
-
-                                Maxim A. Yurkin
-
-                 Institute of Chemical Kinetics and Combustion,
-               Siberian Branch of the Russian Academy of Sciences,
-                  Institutskaya 3, Novosibirsk, 630090, Russia,
-                    tel: +7-383-333-3240, fax: +7-383-334-2350
-
-                               Alfons G. Hoekstra
-
-               Faculty of Science, Section Computational Science,
-                       of the University of Amsterdam,
-              Kruislaan 403, 1098 SJ, Amsterdam, The Netherlands,
-                   tel: +31-20-525-7530, fax: +31-20-525-7490
-
-
-                           email: [email protected]
-
-                     $Date::                            $
-
-                Copyright (C) 2006-2008 University of Amsterdam
-        This software package is covered by the GNU General Public License.
-
-
-
                                 1. INTRODUCTION
                                 ***************
 
@@ -77,7 +76,6 @@ format. It contains instructions for:
 * specifying what scattering quantities should be calculated;
 * understanding the computational aspects and timing of the code;
 * understanding the command line options and formats of input and output files.
-* modifying the source code for added functionality
 
 
                                 2. AVAILABILITY
@@ -87,10 +85,8 @@ format. It contains instructions for:
 it will prove to be a useful tool.  We ask only that:
 
 * If you publish results obtained using ADDA, you should acknowledge the source
-  of the code. We recommend the following general reference:
-  M. A. Yurkin, V. P. Maltsev, and A.G. Hoekstra, "The discrete dipole
-  approximation for simulation of light scattering by particles much larger than
-  the wavelength", J. Quant. Spectros. Radiat. Transf. 106, 546-557 (2007).
+  of the code. Please contact the authors ([email protected]) for the most
+  recent reference.
 
 * If you discover any errors in the code, please promptly communicate them to
   the authors.
@@ -118,11 +114,9 @@ doc/ - documentation
   copyleft - GNU General Public License
   history - complete history of ADDA development
   faq - frequently asked questions
-  manual.doc - source of User Manual in MS Word format
-  manual.pdf - user Manual for ADDA in PDF format
+  manual.doc - Source of User Manual in MS Word format
+  manual.pdf - User Manual for ADDA in PDF format
   README - this file
-  todo.xls - source of the todo list in MS Excel format
-  todo.pdf - todo list in PDF format
 input/ - default input files
   tables/ - 10 auxiliary files with tables of integrals
   alldir_params.dat - parameters for integral scattering quantities
@@ -139,8 +133,8 @@ src/
   Makefile, make_seq, make_mpi - makefiles
   ADDAmain.c, CalculateE.c, calculator.c, cmplx.h, const.h, crosssec.c/h,
   comm.c/h, debug.c/h, fft.c, function.h, GenerateB.c, io.c/h, iterative.c,
-  make_particle.c, matvec.c, memory.c/h, os.h, param.c/h, parbas.h,
-  prec_time.c/h, Romberg.c/h, sinint.c, timing.c/h, types.h, vars.c/h
+  make_particle.c, matvec.c, memory.c/h, os.h, param.c/h, prec_time.c/h,
+  Romberg.c/h, timing.c/h, types.h, vars.c/h
             - source and header files of ADDA
   cfft99D.f - source file for Temperton FFT
   mt19937ar.c/h - source and header files for Mersenne Twister random generator.

doc/faq

@@ -1,7 +1,7 @@
                         Frequently Asked Questions
-                              about ADDA
+                            about Amsterdam DDA
 
-                   $Date::                            $
+                        last revised: 3 June 2007
 
 Q: I have found a bug in ADDA. What should I do?
 A: 1) Make sure you are using the latest version of ADDA (check the ADDA
@@ -14,7 +14,6 @@ A: 1) Make sure you are using the latest version of ADDA (check the ADDA
    parameters as possible without removing the bug. Also try to use defautl
    versions of input files, that you have modified.
    5) Send the results of (3) and (4) to the authors, together with all input
-
    files and Makefiles that you have used for compilation. Do not forget to
    include all the relevant output files,  at least 'log'. Please also include
    a brief description of your operation system and hardware. We will try to
@@ -71,14 +70,5 @@ A: The simplest is to specify your particle by a shape file. However, if your
    source files to the authors so they would be incorporated in the next
    release for the benefit of the community.
 
-Q: How is the Mueller matrix, produced by ADDA, defined and/or normalized?
-A: It is defined as in Bohren & Huffman "Absorption and scattering of Light by
-   Small Particles" (1983), and it is not normalized. Some other codes may
-   compute Stokes scattering matrix, which is normalized so that 1,1-element
-   is equal to 1 after averaging over the whole solid angle. This matrix
-   should be multiplied by (pi*Csca/(lambda^2)) to get Mueller matrix. Csca is
-   the scattering cross section for unpolarized light, equal to average of
-   scattering cross sections for any two perpendicular incident polarizations.
-
 This list is far from being complete. Please send your questions to
 [email protected]

doc/history

@@ -15,7 +15,7 @@ Ver. 0.74 - 29.03.06
 + User manual improved.
 * 'buggy' beam type removed (obsolete).
 * '-nosym' and '-sym_enf' options combined to '-sym {no|enf}'.
-- In documentation there was option '-pol ldr avg' while actually it was
+- In documentation there was option  '-pol ldr avg' while actually it was
   '-pol ldr avgpol'. It is averaging over incident polarizations
   (not directions). Documentation corrected.
 + Internal fields are now automatically collected into one file, when running
@@ -96,7 +96,7 @@ Ver. 0.76 - 14.10.06
 + Makefiles modified to include new compiler 'compaq'. New variables added to
   set path for FFTW3 header and library.
 * Default dpl is now calculated based on the maximum absolute value among given
-  refractive indices, not the first one. dpl=10*max(|m|)
+  refractive indices, not the first one. dpl=10*sqrt(max(|m|))
 - Removed requirement of full set of refractive indices to be given for multi-
   domain scatterers, if '-prognose' option is used. Produces information
   message when relevant.
@@ -183,94 +183,3 @@ Ver. 0.77 - (05.06.07)
   discuss in detail multi-core PCs. Sections "Applicability of DDA" and "System
   requirements" were extended to include recent benchmarking results and
   discussion. Thanks to Vitezslav Karasek and Liviu Clime for their feedback.
-
-----------------------------------
-Ver. 0.78 - (19.03.08)
-
-* Makefiles were improved to enable automatic compilation of both sequential and
-  MPI versions from the same source folder. Name of MPI executable is now
-  adda_mpi.
-+ New file parbas.h added. MPI version is now checked for conformity both during
-  compilation and at runtime.
-+ Parallel executable for Win32, compiled with MPICH 2, is included in the
-  package.
-+ sample/test.sge file was updated to run on DAS-3 cluster.
-+ Added explicit description of the used notation for the Euler angles in the
-  manual, source code, help system, and parameter file.
-+ sample/test.pbs was slightly improved.
-+ File sinint.c added to calculate sine and cosine integrals.
-+ Function cEqual was added to cmplx.h, it replaces memcpy calls in many other
-  files.
-+ Filtered coupled dipoles implemented (both full and quasistatic version). New
-  arguments 'fcd' and 'fcd_st' for interaction term ('-int' command line option)
-  and 'fcd' for polarization term ('-pol' command line option).
-+ To-do list added (both in xls and pdf format).
-+ Hands-on tutorial was added to the manual. It is based on the one carried out
-  during the DDA Workshop in Bremen.
-- ADDA was crashing for certain incorrect parameters given with the -h command
-  line option. Fixed.
-+ A few explicit statements that angles are specified in degrees were added to
-  the help and input files.
-+ New command line option '-opt {speed|mem}' added. It allows to choose whether
-  ADDA will optimize itself for maximum speed or for minimum memory usage.
-+ Stability of complex arithmetics slightly improved.
-+ Calculation of scattered fields is accelerated by more than 4 times (compiled
-  with gcc). Thanks to R. Scott Brock for this idea. This may also significantly
-  accelerate orientation averaging, depending on the problem. Intel compiler is
-  now only 10-20% faster than gcc, for all parts of the code.
-- Minor bug in symmetry initialization of shape 'spherebox' is fixed.
-- Information messages, when no real dipoles were allocated to one of the
-  processors, did not work. Fixed.
-* Information messages ("INFO:...") are now given without references to source
-  files, which looks simpler.
-+ New command line option '-eq_rad <arg>' was added, it allows one to specify
-  volume-equivalent radius.
-* Format of shape definitions in 'make_particle.c' was slightly changed.
-* Shapes 'box' and 'spherebox' no longer depend on the second and third argument
-  of '-grid' option.
-+ Shape 'box' now accepts two optional arguments (y/x and z/x aspect ratios)
-  defining a rectangular parallelepiped.
-+ Automatic line wrapping was added for most output to stdout and sderr. ADDA
-  tries to get terminal width from COLUMNS environmental variables and, if
-  fails, uses the default value.
-* '-pedantic' option for gcc was removed from Makefile, since it gave warnings
-  about long predefined strings.
-- Shape 'line' was not working correcty in combination with '-jagged' or
-  explicitely specified grid sizes along y- and z-axes. Fixed.
-+ Two new shapes were added: capsule and egg. They were implemented by Daniel
-  Hahn and Richard I. Joseph.
-+ Description of predefined shapes in the manual was improved.
-* Now, if grid size is auto-initialized from default dpl, ADDA ensures that it
-  is not smaller than 16.
-+ Limitation on positiveness of dipole coordinates in shape file was removed.
-  Now ADDA automatically determines the minimum box around the particle and
-  centers it as usual. It also ignores blank lines in the middle or in the end
-  of shape file.
-+ Orientation averaging was optimized. When beta=0 or 180, gamma angle is not
-  relevant but only combinations alpha+-gamma. If alpha is varied in a full
-  range [0,360), only one value of gamma for these specific beta values is
-  calculated, saving a few evaluations of internal fields. Thanks to Antti
-  Penttila for this idea.
-+ Limited two-way compatibility with DDSCAT 6.1 geometry format was added,
-  corresponding to its shape option FRMFIL and output of 'calltarget' utility.
-  ADDA automatically detects DDSCAT format during reading of dipole file. When
-  saving geometry to the file, the format is determined by new command line
-  option '-sg_format {text|text_ext|ddscat}'.
-+ New argument 'auto' was added to '-sym' command line option to correspond to
-  the general rule that one of the possible arguments is the default one. No
-  functionality is changed for other arguments.
-
-----------------------------------
-Ver. 0.78.1 - 08.04.08
-
-- Critical bug in cDiv and cDivSelf functions in source file cmplx.h was fixed
-  (introduced in version 0.78). This bug made BiCGSTAB and BiCG iterative
-  solvers to fail. However, even when other two iterative solvers were used,
-  erroneous results were calculated for certain values of refractive index.
-
-----------------------------------
-Ver. 0.78.2 - 11.04.08
-
-- An implementation of a new FCD polarization prescription (-pol fcd) was
-  somewhat faulty. Correction terms of orders (kd)^2 and ln(...)(kd)^3 were
-  interchanged. Fixed.

input/alldir_params.dat

@@ -27,7 +27,6 @@ eps=0
 equiv=true
 periodic=true
 
-# all angles are specified in degrees
 # all values are precalculated; so high 'eps' does not decrease computational 
 # time, but may decrease accuracy. If eps=0, Jmin is not used.
 

input/avg_params.dat

@@ -3,16 +3,12 @@
 # This file should be manually modified by user. 
 # Program does not assume any symmetries of the particle.
 # Therefore, possible symmetries should be considered by user and this can lead
-# to decrease of integration limits. 
-# Here zyz-notation (or y-convention) is used for the Euler angles.
+# to decrease of integration limits 
 
 alpha:
 # calculation for alpha is cheap but only precalculated, therefore
 # Jmax should be rather large.
 # Jmin and eps are really not used;
-# Do not change the range from default unless you have a good reason;
-#   using only one value of gamma for beta=0,pi is done only when full
-#   range is specified here to avoid possible inaccuracies.
 # default: min=0;max=360;Jmax=5;equiv=true;periodic=true
 min=0
 max=360
@@ -47,7 +43,6 @@ eps=1e-3
 equiv=true
 periodic=true
 
-# all angles are specified in degrees
 # Jmin,Jmax are minimum and maximum numbers of refinement stages
 # Nmax = 2^Jmax + 1
 # for those with equiv=true Nmax is effectively less by 1 

input/scat_params.dat

@@ -53,7 +53,6 @@ eps=0
 equiv=true
 periodic=true
 
-# all angles are specified in degrees
 # all values are precalculated; so high 'eps' does not decrease computational 
 # time, but may decrease accuracy
 # Jmin,Jmax are minimum and maximum numbers of refinement stages

sample/run000_sphere_g16m1_5/log

@@ -1,4 +1,4 @@
-Generated by ADDA v.0.78.2
+Generated by ADDA v.0.77
 The program was run on: dda
 command: './adda '
 lambda: 6.283185307
@@ -24,7 +24,6 @@ Scattering quantities formulae: 'by Draine'
 Interaction term prescription: 'as Point dipoles'
 FFT algorithm: FFTW3
 Iterative Method: QMR (complex symmetric)
-Optimization is done for maximum speed
 The FFT grid is: 32x32x32
 Memory usage for MatVec matrices: 1.3 Mb
 Total memory usage: 2.2 Mb
@@ -48,14 +47,14 @@ RE_011 = 2.4111231314E-02  +   progress = 0.193557
 RE_012 = 3.8307716444E-03  +   progress = 0.841121
 RE_013 = 3.0434605426E-03  +   progress = 0.205523
 RE_014 = 1.3101400518E-03  +   progress = 0.569523
-RE_015 = 8.2588869902E-04  +   progress = 0.369618
-RE_016 = 5.0452779281E-04  +   progress = 0.389109
-RE_017 = 1.1339755335E-04  +   progress = 0.775240
-RE_018 = 9.3640503439E-05  +   progress = 0.174228
-RE_019 = 6.8748633527E-05  +   progress = 0.265824
-RE_020 = 2.2386152530E-05  +   progress = 0.674377
-RE_021 = 1.5169835168E-05  +   progress = 0.322356
-RE_022 = 3.1682277704E-06  +   progress = 0.791149
+RE_015 = 8.2588869908E-04  +   progress = 0.369618
+RE_016 = 5.0452779283E-04  +   progress = 0.389109
+RE_017 = 1.1339755286E-04  +   progress = 0.775240
+RE_018 = 9.3640503878E-05  +   progress = 0.174228
+RE_019 = 6.8748635222E-05  +   progress = 0.265824
+RE_020 = 2.2386137724E-05  +   progress = 0.674377
+RE_021 = 1.5169807150E-05  +   progress = 0.322357
+RE_022 = 3.1682119643E-06  +   progress = 0.791150
 
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                 Timing Results             
@@ -66,16 +65,16 @@ Total planes of E field calculation (each 181 points): 2
 Total wall time:     1
 --Everything below is processor times--
 Total time:          0.6300
-  Initialization time: 0.2100
+  Initialization time: 0.2600
     init Dmatrix         0.0100
-    FFT setup:           0.2000
+    FFT setup:           0.2500
     make particle:       0.0000
-  Internal fields:     0.3700
-    one solution:        0.3700
-      init solver:         0.0200
-      one iteration:       0.0200
-  Scattered fields:    0.0300
-    one plane:           0.0200
-  Other sc.quantities: 0.0100
-  File I/O:            0.0100
+  Internal fields:     0.2600
+    one solution:        0.2600
+      init solver:         0.0100
+      one iteration:       0.0100
+  Scattered fields:    0.1100
+    one plane:           0.0500
+  Other sc.quantities: 0.0000
+  File I/O:            0.0000
 Integration:         0.0000